Role: AI Architect
Location: London, UK
(Exp Range 12+)
Type of work: Contract
Mode of work: Hybrid 2 days Onsite in week
JD:
Essential skills/knowledge/experience:
* Design agentic AI architectures using multi‑agent orchestration patterns (planner‑executor, supervisor‑worker, tool‑using agents).
* Define reference architectures for enterprise agent platforms integrating LLMs with systems of record (core banking, CRM, risk, payments).
* Select and standardize frameworks (e.g., LangGraph, Google ADK, MCP, A2A patterns).
* Hands‑on expertise with agentic frameworks (orchestrators).
* Experience with LLMs, prompt engineering, tool/function calling, memory management.
* Exposure to AI quality metrics: task success rate, groundedness, containment, FCR.
* Experience on Google Cloud Platform (preferred) or equivalent hyperscale.
* Deep understanding of LLMs, generative AI, RAG patterns, vector databases, embeddings, and prompt/guardrail engineering.
Desirable skills/knowledge/experience:
* Build Production Grade Agentic AI applications and necessarily have navigated through security, governance and performance management of the end to end model.
* Hands-on expertise on MLOps / AgentOps, CI/CD, and observability.
* Strong understanding of regulated financial services environments.
* Proven experience implementing AI risk controls, model governance, and auditability.
* Ensure alignment with FCA, PRA, data privacy, model risk management, and LBG internal policies.
* Knowledge of banking processes: retail banking, lending, payments, compliance.
* Strong experience designing end‑to‑end AI/ML architectures, including data ingestion, feature engineering, model training, deployment, and monitoring.
* Hands‑on expertise with cloud AI platforms (GCP).
* Strong knowledge of LLMOps practices, including CI/CD for models, model registries, feature stores, lineage tracking, and observability.
* Experience architecting scalable real‑time, batch, and streaming inference systems with clear performance and reliability requirements.
* Familiarity with secure cloud patterns, IAM, VPC design, and data protection
